Formulating Regional Typologies

CIFOR-ICRAF Indonesia

Thursday Nov 2, 2023

Agenda

  1. Background Context
  2. Understanding the Regional Typologies
  3. Methodology
  4. Case Studies
  5. Closing Remarks

Application of Typology in the Development Masterplan for Integrated Land and Seascape

  • The development of an integrated land and sea masterplan (green economic growth) is structured around planning units.

  • Each planning unit is defined by a range of specific characteristics, derived from multiple layers of spatial data.

  • The concept of “Regional Typology” identifies common characteristics—whether in villages, districts, or special units—that guide the formulation of both planning units and strategy or intervention recommendations.

  • The planning process involves a continuous flow of data and information from macro to meso to micro levels, and vice versa.

  • The masterplan will include a range of tailored strategies and on-site interventions, reflecting the specific needs of each planning unit.

What are Regional Typologies?

  • Definition
    • A categorization of geographical areas based on shared features.
    • Features often include quantifiable socio-economic indicators and environmental factors.

Why are Regional Typologies Useful?

  • Regions vary in socio-economic and environmental aspects.
  • Typologies group regions with similar characteristics, making them more manageable.
  • They facilitate comparisons and targeted management.

How Do Typologies and Indexes Differ?

Typology

Urban-rural typology for NUTS level 3 regions

Index

Global Food Security Index

How Do Typologies and Indexes Differ?

Criteria Regional Typologies Regional Indexes
What it is Categorises areas based on similar socio-economic, environmental, or institutional characteristics. A single, composite measure that quantifies the level of vulnerability to a threat.
Outcome Classifications like “urban area” or “rural area.” Numerical score ranking areas from low to high vulnerability.
Use-case Tailoring policies or interventions to different types of areas. Prioritising aid, resources, or interventions based on the ranking
Caveats Can oversimplify complexities; subject to analyst bias. May not capture the full situation; choice of indicators and weightings can be arbitrary.
Complementary aspects Provides rich, contextual information. Pinpoints areas needing immediate attention.

How Do We Make Regional Typologies?

  • Formulate Objectives
  • Identify Data requirement and availability
  • Data collection & pre-processing
  • Variable selection and PCA1
  • Cluster analysis2 and validation
  • Interpretation and comparisons

Case Studies

Case Study #1

Assessing and Characterizing Climate Vulnerabilities in Agriculture-Based Livelihoods: South Sulawesi

Aim: Analyze vulnerabilities of agriculture livelihoods at the provincial level; identify key risks, root causes, and adaptation strategies.

Approach: Define locations with similar biophysical and socio-economic characteristics as ‘typologies’ using PCA & K-means clustering.

Area of Interest: South Sulawesi province

Unit of Analysis: Sub-district (Kecamatan)

No Feature Source Unit
1 Distance to plantation Land cover map 2020, MoEF m
2 Distance to road BIG m
3 Distance to commodity processing factory ICRAF (2016) m
4 Distance to plantation concession South Sulawesi Government m
5 Distance to forest Land cover map 2020, MoEF m
6 Distance to river BIG m
7 Distance to burned area MoEF, 2019 m
8 Percentage of agricultural area (small holder) Land cover map 2020, MoEF %
9 Percentage of plantation area per sub-district Land cover map 2020, MoEF %
10 Percentage of forested area in the sub-district Land cover map 2020, MoEF %
11 Percentage of shrubland in the sub-district Land cover map 2020, MoEF %
12 Percentage of water area compared to sub-district area Land cover map 2020, MoEF %
13 Distance to deforestation Land cover map 2020, MoEF m
14 Deforestation area size Land cover map 2020, MoEF km2
15 Arable land Land cover map 2020, MoEF %
16 Erosion RUSLE t ha-1 yr-1
17 Flood hazard index RBI BNPB index value
18 Land slide hazard index RBI BNPB index value
19 Aridity index WORLDCLIM 2.1 index value
20 Number of Households Potensi desa BPS 2019 number
21 Electrification ratio Potensi desa BPS 2019 %
22 High school Potensi desa BPS 2019 number
23 University Potensi desa BPS 2019 number
24 Hospital Potensi desa BPS 2019 number
25 Health facility Potensi desa BPS 2019 number
26 Market Potensi desa BPS 2019 number
27 Minimarket Potensi desa BPS 2019 number
28 Landslide events 2018-2019 Potensi desa BPS 2019 events/year
29 Landslide fatalities 2018-2019 Potensi desa BPS 2019 people/year
30 Flood events 2018-2019 Potensi desa BPS 2019 events/year
31 Flood fatalities 2018-2019 Potensi desa BPS 2019 people/year
32 Flash flood events 2018-2019 Potensi desa BPS 2019 events/year
33 Flash flood fatalities 2018-2019 Potensi desa BPS 2019 people/year
34 Land and forest fire events 2018-2019 Potensi desa BPS 2019 events/year
35 Land and forest fire fatalities 2018-2019 Potensi desa BPS 2019 people/year
36 Land drought events 2018-2019 Potensi desa BPS 2019 events/year
37 Land drought fatalities 2018-2019 Potensi desa BPS 2019 people/year
38 Natural disaster early warning systems Potensi desa BPS 2019 Number
39 Percentage natural disaster early warning systems Potensi desa BPS 2019 %
40 Reservoir Potensi desa BPS 2019 number
41 Village markets Potensi desa BPS 2019 number
42 Number of people suffering from malnutrition 2018 Potensi desa BPS 2019 individuals
43 Annual mean temperature WORLDCLIM 2.1 °C
44 Temperature changes WORLDCLIM 2.1 & MRI-ESM2-0 SSP 245 2050s °C
45 Annual mean precipitation WORLDCLIM 2.1 mm
46 Precipitation changes WORLDCLIM 2.1 & MRI-ESM2-0 SSP 245 2050s mm
47 Households in the lowest 40% of the economic bracket National Team for the Acceleration of Poverty Reduction %
48 Distance to irrigated land Ministry of Public Works and Housing m
49 Percentage of irrigated land Ministry of Public Works and Housing %
50 Wet months WORLDCLIM 2.1 months
A sample of the Data
ID District Sub-district Dist. to Plantation (m) Dist. to Road (m) Dist. to Comm.Proc Factory (m)
7301010a KEPULAUAN SELAYAR PASIMARANNU 221,140.99 3,151.66 295,618.47
7301011a KEPULAUAN SELAYAR PASILAMBENA 268,465.24 1,996.38 355,496.81
7301020a KEPULAUAN SELAYAR PASIMASSUNGGU 184,133.49 1,649.59 251,716.26
7301021a KEPULAUAN SELAYAR TAKABONERATE 163,497.97 4,252.98 242,261.12
7301022a KEPULAUAN SELAYAR PASIMASSUNGGU TIMUR 188,167.82 861.54 258,492.26
7301030a KEPULAUAN SELAYAR BONTOSIKUYU 99,706.86 744.58 173,876.77
Principal Component Analysis Summary
Importance of components
Components PC1 PC2 PC3 PC4 PC5 PC6 PC7 PC8 PC9 PC10
Standard deviation 2.90 2.22 2.03 1.76 1.64 1.44 1.36 1.32 1.16 1.12
Proportion of Variance 0.17 0.10 0.08 0.06 0.05 0.04 0.04 0.03 0.03 0.03
Cumulative Proportion 0.17 0.27 0.35 0.41 0.47 0.51 0.54 0.58 0.61 0.63

Loadings Interpretation

PC1: Predominantly Demography & Disaster Risk

PC2: Predominantly Climate Features

PC3: Predominantly Distance Features

The first three principal components, colored by cluster

Cluster Validation

The “elbow” of the plot is the point where adding more clusters doesn’t provide much better fit to the data.

A silhouette plot closer to +1 indicates good clustering, while values closer to 0 or negative values suggest overlapping or poor clustering

Case Study #1

Assessing and Characterizing Climate Vulnerabilities in Agriculture-Based Livelihoods: South Sulawesi

  • The second lowest household density after rural areas.

  • Limited accessibility to land-based livelihoods.

  • Economy predominantly relies on marine resources and maritime transport.

  • Inadequate health, education facilities, and market access.

  • Driest area in South Sulawesi with potential freshwater scarcity.

  • High drought indices with significant predicted rainfall decrease.

  • Over 40% of the population is underprivileged.

  • Mostly situated in warm lowlands; a 1°C temperature rise could be impactful.

  • Rural areas in South Sulawesi with the lowest household density.

  • Mountainous regions overlapping with protected areas, last bastions of unique Wallacean biodiversity.

  • High rainfall, low drought vulnerability (AI = 0.3).

  • In close proximity to forests, deforestation sites, and areas previously affected by fires.

  • Furthest from plantation concessions and limited road and river access.

  • Minimal changes in rainfall but the highest rate of temperature increase.

  • Highest erosion risk; lowest flood risk.

  • Almost 50% of the population falls in the bottom 40% economic bracket.

  • Close to plantations, forests, and deforestation sites.

  • Furthest from previously burned areas in Sulawesi.

  • Medium household density per sub-district.

  • Moderate flood risk.

  • Moderate extent of irrigation.

  • Second highest risk of erosion and landslides after inland areas.

  • Highest rate of deforestation.

  • High risk of rainfall reduction, though not classified as arid.

  • About 1/3 of the population falls in the bottom 40% economic bracket.

  • Highest arable land coverage, with predominant irrigated areas.

  • Highest smallholder land ownership.

  • Second closest proximity to burned areas after urban regions.

  • Second highest access to roads and rivers following urban areas.

  • Second highest household density per district after urban regions; similar trend observed in health and education facilities.

  • Least proximity to forests due to limited forest cover.

  • Primarily located in lowlands; a temperature rise of 1.41°C significantly impacts this already warm region.

  • Expected to experience the most significant rainfall decrease (-84.31 mm) among typologies, making it the most drought-prone.

  • Flood risks during the rainy season, aggravated by minimal remaining forest cover.

  • High household density, with over a third belonging to the bottom 40% economic bracket.

  • Smallest area coverage (3%) with least arable land.

  • Lowest smallholder land ownership, following island regions.

  • Second closest proximity to burned areas after semi-urban regions.

  • Best access to roads and rivers.

  • Highest household density per district.

  • Easy access to health facilities, supermarkets, universities, and hospitals. However, each unit caters to a large number of households.

  • Only 15.1% of households fall within the bottom 40% economic bracket.

  • Warmest current temperatures compared to other regions; a rise of 1.43°C significantly impacts underprivileged local residents.

  • High flood risk, given many urban areas are near rivers.

  • Heavily reliant on food supplies from other regions.

Case Study #2

Village-level vulnerability classes of agricultural-based livelihoods to climate change in the studied districts in West Kalimantan Province

1 – Most Vulnerable: Located closest to oil palm plantations and factories, mining areas, and roads; has the largest shrub area per village; furthest from deforestation and has the lowest deforestation rate; has the smallest percentage of forested areas per village; high population.

2 – Highly Vulnerable: Located closest to burnt areas; has the largest percentage of oil palm area per village; situated nearer to oil palm concession areas and rubber factories; has the largest percentage of water bodies (lakes, rivers); has a lower deforestation rate but is located closer to deforestation areas; high population.

3 – Moderately Vulnerable: Located closest to rivers; has a larger percentage of oil palm area per village; situated slightly farther from oil palm companies and mining areas; has a larger percentage of forested and shrub areas per village; medium population.

4 – Less Vulnerable: Closest to deforestation areas and has the highest deforestation rate; located furthest from rivers; most distant from burnt areas; slightly closer to forested areas; slightly farther from oil palm concession areas; low population.

5 – Least Vulnerable: Has the largest percentage of forested areas per village; most remote; has the smallest percentage of shrub areas and oil palm areas per village; located closer to rivers and forested areas; has the lowest village population.

No Category Spatially explicit variables References

Stashed changes | 1 | Distance to infrastructure | Distance to oil palm plantation | Landcover map 2017 from MoEF | | 2 | Distance to infrastructure | Distance to roads | BIG | | 3 | Distance to infrastructure | Distance to a rubber factory | ICRAF | | 4 | Distance to infrastructure | Distance to oil palm factory | ICRAF | | 5 | Distance to natural resources | Distance to forest | Landcover map 2017 from MoEF | | 6 | Distance to natural resources | Distance to river | BIG | | 7 | Distance to natural resources | Distance to mining areas | BIG | | 8 | Distance to hazards area | Distance to burnt areas | KLHK 2015 | | 9 | Land use and land cover | % area of oil palm per village | Landcover map 2017 from MoEF | | 10 | Land use and land cover | % of forested areas per village | Landcover map 2017 from MoEF | | 11 | Land use and land cover | % of shrubs areas per village | Landcover map 2017 from MoEF | | 12 | Land use and land cover | % forested areas per district compared to district areas | Landcover map 2017 from MoEF | | 13 | Land use and land cover | % oil palm areas at district level compare | Landcover map 2017 from MoEF | | 14 | Land use and land cover | % water body compared to district areas | Badan Informasi Geospatial (BIG), Geospatial Information Agency of Indonesia | | 15 | Land use and land cover | Distance to deforestation | Landcover map 2017 from MoEF | | 16 | Land use and land cover | Deforestation area | Landcover map 2017 from MoEF | | 17 | Hazards | Flood incidence | Village Potentials, Statistics Indonesia 2018 | | 18 | Hazards | Heavy flood incidence | Village Potentials, Statistics Indonesia 2018 | | 19 | Hazards | Fire incidences | Village Potentials, Statistics Indonesia 2018 | | 20 | Hazards | Drought incidences | Village Potentials, Statistics Indonesia 2018 | | 21 | Demography | Village population | Village Potentials, Statistics Indonesia 2018 | | 22 | Distance to infrastructure | Distance to oil palm concession areas | Department of Estate Crops, West Kalimantan Province |

: {.striped .hover}

Case Study #3

Peat Ecosystem Typologies in South Sumatra Province

Typology 1: High fire occurrences in protected areas, both with and without drainage. Lowest rates of deforestation, degradation, and plantation expansion. Low population and poverty, closest to cities with easy access.

Typology 2: High fire occurrences in cultivated areas, both with and without drainage. High deforestation rate, low degradation, and minimal plantation expansion. Highest population and poverty, near cities with easy access.

Typology 3: Fires occur in both cultivated and protected areas, irrespective of drainage. Highest rates of deforestation, degradation, and plantation expansion. Low population and poverty, furthest from cities with challenging access.

Typology 4: Most frequent fires in protected areas, regardless of drainage. Low deforestation, high degradation, and significant plantation expansion. Low population and poverty, far from cities with very limited access.

Typology 5: Most frequent fires in cultivated areas, irrespective of drainage. Low deforestation, high degradation, and significant plantation expansion. Low population and poverty, remote from cities with challenging access.

Typology R.3.1 Establishment of Peatland Stewardship Village R.3.2 Enhancing Livelihood Capacity R.3.3 Development of Alternative Commodities and Livelihood Sources

Stashed changes | 1 | - Community awareness about the importance of the peat ecosystem, collaborating with district and provincial governments.

- Activities of the Peatland Stewardship Village aimed at increasing income and creating sources of livelihood for the community. | Conducted in synergy with district/city governments. | Directed at non-agricultural livelihood sources (off-farm). Local industries that exist need support. | | 2 | Peatland Stewardship Village is executed in an integrated manner in partnership with HTI companies and palm oil plantations. The goal is to enhance community access. | Strengthening farmers’ livelihoods to protect the remaining peat ecosystems. Formulation of alternative livelihood sources. | Livelihood sources oriented towards peat-friendly agriculture, mixed plantations combined with non-agricultural livelihood sources. | | 3 | Peatland Stewardship Village is oriented towards fire prevention; conducted in an integrated manner in partnership with HTI companies and palm oil plantations. | Strengthening farmers’ livelihoods through partnerships with plantation and HTI companies. Livelihoods aimed at forest and land fire prevention. | Livelihood sources oriented towards peat-friendly agriculture and plantations combined with fisheries and livestock. | | 4 | Peatland Stewardship Village geared towards fire prevention; enhancing community access to markets and high-economic value commodities. | Strengthening farmers’ livelihoods through partnerships with plantation and HTI companies. Emphasis on preventing forest and land fires. | Livelihood sources focused on peat-friendly agriculture and plantations combined with fisheries and livestock. | | 5 | Peatland Stewardship Village focused on fire prevention; enhancing community access to markets and commodities of high economic value. | Strengthening farmers’ livelihoods through partnerships with plantation and HTI companies. Emphasis on preventing forest and land fires. | Livelihood sources focused on peat-friendly agriculture and plantations combined with fisheries and livestock. |

: Classification of peat ecosystem revitalization actions in South Sumatra province {.striped .hover}

Conclusion

  • We showcase examples on how regional typologies can be produced
  • The aim of the study shapes the data selection and interpretation of the typologies.
  • Iterative feedback from experts is essential for ensuring typologies align with objectives.

Footnotes

  1. Principle Component Analysis is a statistical technique used to simplify the complexity in data by highlighting its most important features.

  2. Cluster analysis is a machine learning procedure used to group similar items together based on their characteristics.